In general, eye-to-eye contact in a conversation between people is an important factor for enabling one person to recognize how much other people are concentrating on his/her talk. The eye-to-eye contact is a tacit language for enabling a person to be aware of a response and emotion, which are not said, by another person, and there is a research result that reliability and a sense of closeness to a person having an eye-to-eye contact are increased. Accordingly, eye-to-eye contact between remote participants in a conference using the teleconference system has attracted attention as one of a core technique of improving participation in the conference.
In general, in the teleconference system, a camera is installed and operated on a screen (or a monitor) as illustrated in FIG. 1. Accordingly, even though a local participant looks at a face of a remote participant shown on the screen, the camera is located on the screen, so that the eyes of the local participant look downwardly in an image obtained by the camera, which is shown to the remote participant, by an angle difference θ generated for the local participant looking at the screen. Accordingly, even though the local participant actually looks at the eyes (or the face) of the remote participant shown on the screen, but the remote participant feels that the local participant is looking downwardly, not at his/her eyes (or the face). By contrast, when the camera is located at a lower side of the screen, the eyes of the local participant look upwardly in the image obtained by the camera even though the local participant looks at the remote participant shown on the screen.
In order to solve the problem, many organizations have conducted research on eye-to-eye contact technology. The eye-to-eye contact technology for the teleconference system is generally divided into two methods, that is, a physical method and an image combination method, and research thereof has been conducted.
The physical method means technology for providing an eye-to-eye contact image by adjusting a position of a camera. In general, an image is obtained by positioning a camera at a center on a back side of a semi-transmissive screen or adjusting a position of the camera to an eye-level of a local participant as close as possible to provide eye-to-eye contact. In the physical method, the former case requires installation of a separate system, and the latter case has a problem in that the eyes of a local participant may be hidden.
The image combination method means technology for combining eye-to-eye contact images by using images obtained by one or more cameras. A representative method is a method of pre-storing an image for eye-to-eye contact and replacing only an eye part with the eye-to-eye contact image in an image obtained by a camera, or combining 3-dimensional images by an image processing method, such as stereo matching, converting an eye point into a front side to provide eye-to-eye contact. In this method, it is very important to naturally combine eye-to-eye contact images in real time. Further, in applying the eye-to-eye contact technology, when a local participant looks at another point, not the eyes (or a face) of a remote participant, and an eye-to-eye contact is applied, very unnatural images are combined.